Lara A Harvey1, Roslyn G Poulos, Caroline F Finch. 1. NSW Injury Risk Management Research Centre, University of New South Wales, Sydney, NSW, Australia. Electronic address: l.harvey@neura.edu.au.
Abstract
UNLABELLED: Routinely collected hospitalisation data are widely used to monitor injury trends, provide estimates of the burden of injury and healthcare costs, and to inform policy. This study examined the impact of different ICD-10 based case selection criteria commonly used by Australian and international reporting bodies on the number and nature of burn-related hospitalisations identified. METHODS: Burn cases from a state-wide administrative hospitalisation dataset were identified and compared using three different case selection criteria: (1) principal diagnosis code of burn 'T20-T31', (2) first external cause code denoting burn 'X00-X19' and (3) both principal diagnosis code of community acquired injury 'S00-T98' and first external cause code denoting burn 'X00-X19'. RESULTS: Principal diagnosis codes 'T20-T31' and first external cause codes 'X00-X19' identified a similar number of cases, however only 78% of these were captured by both definitions. Principal diagnosis codes identified chemical, electrical and contact burns not identified as burns using external cause codes. First external cause codes identified readmission cases which were not identified by principal diagnosis codes. Using principal diagnosis codes of community acquired injury combined with external cause code of burn under-numerated hospitalisations by forty percent. CONCLUSION: The development, implementation and evaluation of health policy and prevention measures rely on good quality, consistent data. Current methods for identifying burn cases in hospitalisation data provide wide differences in estimation of number and nature of cases. It is important for clinicians to understand the implications of coding on the epidemiology and measurement of the burden of burn.
UNLABELLED: Routinely collected hospitalisation data are widely used to monitor injury trends, provide estimates of the burden of injury and healthcare costs, and to inform policy. This study examined the impact of different ICD-10 based case selection criteria commonly used by Australian and international reporting bodies on the number and nature of burn-related hospitalisations identified. METHODS: Burn cases from a state-wide administrative hospitalisation dataset were identified and compared using three different case selection criteria: (1) principal diagnosis code of burn 'T20-T31', (2) first external cause code denoting burn 'X00-X19' and (3) both principal diagnosis code of community acquired injury 'S00-T98' and first external cause code denoting burn 'X00-X19'. RESULTS: Principal diagnosis codes 'T20-T31' and first external cause codes 'X00-X19' identified a similar number of cases, however only 78% of these were captured by both definitions. Principal diagnosis codes identified chemical, electrical and contact burns not identified as burns using external cause codes. First external cause codes identified readmission cases which were not identified by principal diagnosis codes. Using principal diagnosis codes of community acquired injury combined with external cause code of burn under-numerated hospitalisations by forty percent. CONCLUSION: The development, implementation and evaluation of health policy and prevention measures rely on good quality, consistent data. Current methods for identifying burn cases in hospitalisation data provide wide differences in estimation of number and nature of cases. It is important for clinicians to understand the implications of coding on the epidemiology and measurement of the burden of burn.